Stephen Downes

On the way to the office in what was an overly dangerous commute this morning I listed to the TWIT Intelligent Machines interview with Pliny the Liberator, an anonymous prompt-writer known for jail-breaking AI engines. This link is to the GitHub repository of scripts that break through the guardrails on services offered by Anthropic, Grok, and others. If you're interested, do listen to the interview (or read the transcript) comprising the first 37 minutes of this podcast episode. It poses the question: do AI guardrails constitute a form of censorship not only on the AI but also on the people who use them? If so, then what role should the government, educational institutions, and the public at large play in the development and transparency of these guardrails?